The secular decline in teen employment: the role of compulsory

schooling and work permitsPreliminary. Please do not cite without author’s permission ∗

Ciprian Domnisoru†

Carnegie Mellon University

April 15, 2015

Abstract

I analyze the role that compulsory schooling and employment certificate legislation play in

the secular decline in teenage employment since 2000. Using a difference in differences strategy

and Current Population Survey data, I show that increases in school leaving ages in 20 states

since 1985 have reduced the employment rate of 16 and 17-year-olds by one to two percentage

points. Using a similar empirical strategy, I analyze the effect of dropping work permit require-

ments for 16 and 17-year-olds in five states that did so between 1977 and 1996. The employment

rate of 16 and 17 year-olds increased by about two percentage points after employment certifi-

cates were no longer required.

JEL Keywords: I20, I21, I28, J24, J31

∗The author would like to thank his advisers, Karen Clay, Brian Kovak and Lowell Taylor, and his colleagues

participating in the Heinz College PhD seminars, for their suggestions and comments†Ciprian Domnisoru, Ph.D. Student, H.John Heinz III College, Carnegie Mellon University. Email: cdom-

niso@andrew.cmu.edu

1

1 INTRODUCTION

1 Introduction

Teenage employment has experienced a secular decline since 2000. The decline was triggered by

the March-November 2001 recession, but continued through stages of expansion and accelerated

in the latest recession (Figure 1). Analyses of this secular decline have argued that business cycle

patterns cannot fully acccount for the drop in teenage employment (Aaronson et al, 2006; Smith,

2011). Using data until the end of 2010, Smith (2011) showed the counterfactual employment to

population rate was two percentage points above the actual rate. The literature has not, however,

analyzed the role played by increases in the maximal school leaving age introduced in 20 states since

1987, nor the increasingly tougher truancy and employment certificate laws. I show that increases

in the school leaving age account for one to two percentage points of the decrease in employment in

states that passed compulsory schooling laws. The 20 states passing such laws account for 44 per

cent of the 2013 US population. The employment decline in these states following the introduction

of compulsory schooling laws accounts for about one percentage point of the unaccounted decline

in the employment to population ratio.

Working during high school is increasingly less common in the United States. In the 1980 High

School and Beyond Survey, 44 per cent of enrolled sophomores and 64 per cent of enrolled seniors

were employed during the school year (Lewin-Epstein, 1981). Figure 1 shows the significant decline

in employment rates for 16 and 17-year-olds between 1984 and 2013 : from 35 to about 10 per cent

for seventeen year olds, and from 35 to under 20 per cent for seventeen year olds. Aaronson et

al (2006) and Smith (2012) have considered increases in school enrollment as explanations for the

decline in teenage employment. Figure 2 however shows that the enrollment rate of 16-year-olds

has been relatively constant between 1994 and 2008, a period in which employment decreased con-

siderably. Enrollment increases for seventeen year olds have not exceeded two percentage points

since 1994, while the drop in employment was about 15 percentage points. The decline in labor

force participation over the past ten years has been so significant that it impacts overall labor

force participation. As such, it became a puzzle, leading researchers to discuss the unexplained

nature of part of the decline (Aaronson et al, 2006) and to conclude ”there is little research on why

employment has fallen” (Smith, 2011).

But apart from explaining the macroeconomic labor force participation puzzle, the decline in

teenage employment is of policy interest in itself. Part time jobs may give young people better work

habits and help them make more informed career choices. Early labor market experiences may also

raise subsequent earnings. The federal government supports the notion that students should work

during high school as a way of ensuring a smoother transition to full time employment. For exam-

ple, the Federal School-to-Work Transition Act of 1994 provided 100 million dollars in support of

2

1 INTRODUCTION

state school to work initiatives. Working during high school is believed to result in the acquisition

of skills, experience and work habits, which would improve students’ productivity and later chances

of finding employment.

Critics of student employment during high school point to the displacement effect that time

spent working has on time spent learning. In response to these concerns, some states passed bills

to reduce the number of hours high school students are allowed to work during the school week.

Concerns that long hours of work lead to poorer academic outcomes are supported by research.

Tyler (2003) uses NELS88 data and shows that a decrease of 10 hours per week in working hours

results in a 0.2 standard deviation increase in math scores for twelfth graders. Eckstein and Wolpin

(1998) use NLSY79 data and estimate a sequential decision model of high school attendance and

work. They find that working full time results in a 0.135 point reduction in cumulative grade point

average (on a four-point scale).

If employment during high school is to some extent detrimental to academic performance, it

conflicts with another popular policy goal, increasing high school enrollment and graduation rates.

Having a high school diploma significantly reduces the risk of unemployment and leads to higher

lifetime earnings. The policy goal of increasing high school graduation rates has been specifically

addressed through changes in the minimal school leaving age. Twenty states have increased the

minimal school leaving age over the past 25 years (twice, in the case of Louisiana) (Table 1). How-

ever, enforcing compulsory schooling laws when the enrollment rate is more than 90 per cent is a

difficult task, as the reasons why a select, relatively small set of students drop out of high school

may be very diverse, pressing and stronger than any incentives and enforcement mechanisms used

by school truancy boards. I show the effects of such policy changes on the overall enrollment rate

are statistically insignificant, but can reach statistically significant effects of 3 percentage points

for teenagers in households in the lowest household income decile (Table 5).

In the push to enforce compulsory schooling policies, states have resorted to mechanisms that

limit the probable high school drop out student’s outside options: limiting driving priviliges (’no

pass no drive policies’) or conditioning the issuance of a work permit on school enrollment (and

performance) . Work permits, or employment certificates, are child labor requirements meant to

ensure employers document the age of minors and are made aware of the special nightwork, total

hours, health, safety and other special restrictions that apply to minor employees. Eighteen states

currently require an employment certificate for minors aged 16 or 17. When compulsory schooling

laws increase the minimal school leaving age past the age of sixteen, employment certificates can

only be issued if the minor is enrolled in school (with some exceptions, discussed in Section 3).

Section 4 elaborates on a theoretical prediction that compulsory schooling laws send a negative

3

2 RELATION TO EXISTING LITERATURE

signal to employers about the variance of skills and abilities of cohorts impacted by the policy

change: in a nutshell, employers can no longer distinguish between teenagers willingly staying in

school and teenagers compelled to enroll. In states requiring employment certificates, compulsory

schooling also make hiring sixteen and seventeen year olds more costly and difficult because of

work permit fees and extra bureaucratic pressures to comply with certificate requirements, leading

to higher decreases in employment for high school students. I show compulsory schooling laws

requiring 17-year-olds to stay in school until 18 are associated with a two percentage point decrease

in employment.

Legislators in six states relaxed employment certificate requirements between 1973 and 1995, in

an attempt to facilitate the hiring of teenage labor. I show that dropping employment certification

requirements for 16 and 17 year olds was associated with a statistically significant increase of two

percentage points in employment.

Policymakers, school counselors and teachers, parents and students face a dilemma: focus on

academic achievement may increase chances of graduation, but could come at the cost of the positive

effects of teenage employment. Compulsory schooling laws coupled with employment certificates

may lead to modest increases in enrollment, particularly for students from low income households.

But the risk is that while they positively impact a few students to stay in school and perhaps

graduate from high school, such policies may have a larger, negative effect on employment for a

broader set of students than those targeted by compulsory schooling laws.

2 Relation to existing literature

The record lows for teenage employment in recent years have attracted the attention of macroe-

conomists, as the drop has been so significant that it impacts overall labor force participation. The

literature discusses the unexplained decline in labor force participation, after detrending business

cycles. Aaronson, Park and Sullivan (2006) found that if the 1987-1997 rate of decline in teenage

employment had been maintained, the 2006 rate would have been about 5.5 percentage points

higher. However, they find weak evidence of demand factors leading to the decline, as relative

wages of teenagers have not changed substantially. Instead, they find increasing enrollment to be

the main cause of the decline, accounting for 0.18 percentage points per year in the decline of

teenage labor force participation. They do not, however, explore the role of compulsory schooling

laws as drivers of the increase in enrollment. My analysis shows that compulsory schooling affects

employment not just through the sheer increases in enrollment, but have an effect on employment

even under lax or absent enforcement of minimal leaving age increases.

4

2 RELATION TO EXISTING LITERATURE

Smith (2011) explored other potential drivers of the decrease in employment: exit exam re-

quirements , merit aid scolarship programs for college and an increase in the number of credits in

core subjects needed to graduate. These policies put pressure on the labor supply of teenagers as

they increase academic requirements, and, in the case of scolarship programs, displace the need for

students to finance their college tuition by working in high school. Smith finds the effect of such

labor supply explanations modest in comparison to labor demand explanations which stem from

increases in the supply of adult labor that is substitutable by teens, notably unskilled immigrants

(Smith, 2012).

The effects of compulsory laws on educational attainment and later labor market outcomes in

the US have been explored by canonical papers (Angrist and Krueger, 1991; Acemoglu and Angrist,

2001). Using similar methodologies, Oreopoulos (2009) has shown recent reforms to have positive

effects on academic and long term outcomes. Oreopoulos’ analysis follows compulsory schooling

changes only until 2005. It is also affected by errors in the coding of the compulsory schooling

changes that were made in the National center for Education Statistics (NCES) yearly digests.

Whitehurst and Whitfield (2012) and Mackey and Duncan (2013) identified these coding problems

and proceeded to check the actual state statutes and contact state labor departments. Because

of the coding mistakes, esentially all analyses of recent compulsory changes before 2012 may be

significantly biased because of coding errors. Biases may be substantial, as large states such as New

York and Florida were attributed inexistent law changes. Following Mackey and Duncan (2013),

I update the information on actual law changes by proceeding to look for multiple sources and

provide details on the actual statute and law changes (Table 1) 1. I find discrepancies in the coding

used by Whitehurst and Whitfield (2012) and Mackey and Duncan (2013) as well: a 1991 Arkansas

law change increased the compulsory schooling age, mentioning that students aged through 17 on

or before the start of the school year are obliged to enroll. This law change was incorrectly coded

as a maximal school leaving age of 17, while in fact it is a maximal school leaving age of 18.

The effect of employment certificate requirements on teenage employment has received less

attention. The effects of the original implementation of such laws are difficult to analyze, as

employment certificate legislation was introduced at the beginning of the 20th century, and its

enforcement was a gradual process, as state labor inspectorates were just being set up and states

gradually consolidated their legislation on child labor. Tyler (2003) used a snapshot of employment

certificate legislation across US states as an instrument for the labor force participation of high

school students. While there is a correlation between employment certificate legislation and the

employment rate of teenagers (Table 6), Tyler (2003) went no further in exploring the causal links.

1A detailed list of policies and legal references is available from the author upon request

5

3 LAWS AFFECTING TEEN EMPLOYMENT

To my best knowledge, the causal effect of such legislation on employment has not been explored.

I tracked state legislative changes and identified six states that dropped the requirement for em-

ployment certificates for 16 and 17-year-olds between 1973 and 1995: Utah (1973), Nevada(1977),

Virginia (1979), Florida (1981), Kentucky (1984), Oregon (1995). 2 I exploit these law changes

to estimate the effect of dropping work permit requirements on the employment of sixteen and

seventeen year olds.

The rich literature on the effect of minimum wages on teenage employment has acknowledged the

role of compulsory schooling policies. Neumark and Wascher (1995) include compulsory schooling

as covariates. However, they find no statistically significant effect of such policies on employment.

As the minimum wage debate unfolded over the 1990s, compulsory schooling policies were an ”op-

tional extra” in the set of controls. Recent papers challenging established results in the literature

(e.g. Allegreto, Dube and Reich, 2011) or papers revisiting classic results (Neumark, Salas and

Washer, 2013) did not account for compulsory schooling changes in the past twenty years. This

introduces the possibility of bias in estimates of the effects of minimum wage laws. Unaccounted

increases in the minimal school leaving age that occur at the same time as minimum wage increases

may introduce an upward bias in the effect of minimum wage policies if compulsory schooling poli-

cies and minimum wages both negatively impact the employment rate.

3 Laws affecting teen employment

3.1 Compulsory Schooling

Over the past 25 years, 20 states have increased the minimal school leaving age, twice in the case

of Louisiana. In 2014, 20 states and the District of Columbia had a minimal compulsory schooling

leaving age of 18. Twelve other states had 17 as the maximum age, with the rest of states requiring

that students stay in school until their sixteenth birthday. Several states are currently considering

increasing the minimal school leaving age, and, in his 2012 presidential address, President Obama

has encouraged all state legislators to consider raising the school leaving age to 18.

The changes to compulsory schooling legislation in the past 25 years have been part of a na-

tional drive to reduce the high school drop out rates and increase college attendance. Such policies

have generally had the support of educators. For example, The National Association of Secondary

School Principals supports the raising to 18 of the minimum age at which a student is allowed to

2A detailed list of policies and legal references is available from the author upon request

6

3.1 Compulsory Schooling 3 LAWS AFFECTING TEEN EMPLOYMENT

leave compulsory education3 The societal motivation for keeping more students longer in school

is that dropping out early is costly, both to individuals and to society because of earnings losses

and because of higher risks of unemployment and criminal activity. The lifetime earnings of high

school dropouts are considerably lower than those of high school graduates. Compulsory schooling

policies thus have the potential to lead to significant monetary gains if they also induce students

to get a high school diploma.

Raising the compulsory schooling age is enforced through fines or the threat of criminal punish-

ment for parents of truant students. Oreopoulos (2007:91) provides an inventory of such enforce-

ment mechanisms, which include punishment in the form of notices to parents, fines, community

service work, imprisonment, social and rehabilitation service or charges of misdemeanor. Since

1987, I documented 18 states have also begun to condition obtaining a driver’s license on being

enrolled in school. Preliminary evidence (not reported) shows the introduction of no pass no drive

policies does not have a statistically significant effect on the teenage enrollment rate. It remains

,however, one of many enforcement mechanisms.

There are , however, many exceptions that allow students to leave school earlier than the min-

imum required age, either with parental consent or if working. Exceptions include: enrollment

in adult education, parental consent, or employment (in some states only under financial duress).

I currently do not analyze the impact of such exceptions, or the impact of tougher enforcement

mechanisms. Such an analysis would require a detailed review of legislative changes and of state

by state administrative enforcement practices, which may differ in practice from those stipulated

in state laws. It is worth noting, however, that enforcement mechanisms and exceptions have been

changing over the past 20 years. The next section shows an increasing trend in state legislation

to link the issuance of employment certificates for sixteen and seventeen year olds to school enroll-

ment. Teen driving laws that link the issuance of a driving license to enrollment are also recent

developments. Finally, at the beginning of the 1990s, a series of states have proceeded to increase

monetary fines for noncompliance with compulsory schooling and work permit requirements.

Past compulsory schooling policies have been shown to produce earnings gains for students com-

pelled to stay in school . Angrist and Krueger (1991) and Acemoglu and Angrist (2001) estimated

that annual adult earnings were 10 per cent higher for students who had been compelled to stay a

year longer in school by raising the minimal school leaving age. Revisiting the effects of compulsory

schooling policies by analyzing recent changes is warranted by the changing circumstances under

which the newer policies operate. High school drop out rates are lower than during the periods

3National Association of Secondary School Principals , Legislative Advocay,https://www.nassp.org/LegislativeAdvocacy/ [Aug 17, 2013]

7

3.2 Employment certificates for minors 3 LAWS AFFECTING TEEN EMPLOYMENT

analyzed in the literature, and today most students are expected to finish high school. Moreover,

students who drop out today may be facing unique family and environmental circumstances.

3.2 Employment certificates for minors

Under state child labor laws, schools and/or state Labor departments may issue employment cer-

tificates for minors and/or age certification documents. Not all states have such paperwork re-

quirements for the employment of minors. Table 2 shows that twelve states require no employment

certificates, 20 states require an employment certificate for minors under the age of 16 while eigh-

teen states mandate such a certificate until the minor reaches the age of 18. In most states, schools

issue the employment certificate, but in others this may also be done by the State Labor Depart-

ment. Most states that do not require an employment certificate or limit the required age at 16

make the issuance of such a proof for older minors optional. Some states also require employers to

have alternative proofs of the age of minors, such as copies of birth certificates, driving licenses or

a signed note from parents.

Employment certificates for minors serve as instruments in the enforcement of child labor leg-

islation. The employment of minors is prohibited at the federal level in a series of occupations

considered hazardous. To protect the health and physical development of children, the federal gov-

ernment also imposes restrictions on the minimal age at which children may start working, and on

the maximum number of hours they are allowed to work. The restrictions however mostly pertain

to teenagers under the age of sixteen. Some states have more stringent requirements that also cover

the employment of sixteen and seventeen year olds.

Child labor legislation is an active field for state policymakers. Over the past century, most

states consolidated their child labor provisions and increased coverage to older ages or strenght-

ened enforcement mechanisms. However, over the past forty years, some states relaxed child labor

legislation. These states treated the employment certificates as unncecessarily burdensome, while

the former saw them as a valuable tool in the fight against child labor: a 1949 labor legislation con-

ference statement recommended the issuance of employment certificates for minors up to 18 years

of age: ”Prerequisites of the issuance of the certificate shall include a statement of the employer

of his intention to employ the minor, documentary proof of the minors age, school record showing

grade minor has completed, and medical examination showing fitness for the job” (Monthly Labor

Review, Jan 1950:41). In the following paragraphs, I briefly review state legislative changes using

summaries published by the Monthly Labor Review, 1945-2013 . The overall picture is a patchwork

of provisions, in some states relaxing, and in other states strenghtening child labor and employment

8

3.3 Minimum wage laws 3 LAWS AFFECTING TEEN EMPLOYMENT

certificate legislation.

First, some states increased the coverage of employment certificates to sixteen and seventeen

year olds (Kentucky and Virginia, 1948, Delaware, 1951). After 1960, however, most states have

actually proceeded to relax or drop employment certificate requirements for sixteen and seventeen

year olds: Utah (1973), Nevada(1977), Virginia (1979), Florida (1981), Kentucky (1984), Oregon

(1995). Other states have eliminated employment certificates altogether (Arizona, 1972; South

Dakota, 1991; Tennesee, 1978).

Other states have proceeded to relax specific provisions of employment certificate requirements:

In 1970, Virginia made the mandatory physician’s visit valid for two years instead of one year.

Other states eliminated the physical examination requirement for employment certificates alto-

gether (North Carolina, 1971, West Virginia, 1975, Arkansas, 1980). In 1966 in New York, 1969

in Hawaii and in 1973 in North Carolina, teenagers were allowed to use a single permit for all

employers, as opposed to obtaining a different one for each employer. Pennsylvania also introduced

transferable work permits in 1984. State legislators also found that employment certificate legis-

lation was hindering work study programs. In 1970, Massachussetts eliminated the work permit

requirement for students engaged in such programs. States also found permits unncecessary for

high school graduates: California exempted them from such requirements in 1977.

As more states increased compulsory schooling requirements, employment certificate legislation

began to interact with enrollment legislation. In 1989, Maine prohibited the employment of students

habitually truant under the compulsory schooling law, which stipulated 17 as the minimal school

leaving age. Also in 1989, New Hampshire added ’a satisfactory level of academic performance’

to the prerequisites for the issuance of an employment certificate, providing for the possibility of

revocation if the level of academic performance was not maintained. Indiana in 1990 and New York

in 1991 introduced similar provisions, authorizing the revocation of an employment certificate in

case of a significant drop in grades following the issuance of a permit. Out of the 20 states that in-

creased the compulsory schooling age to either 17 or 18, five states (Alabama, Indiana, Louisiana,

California and Michigan) did so while at the same time requiring an employment certificate for

minors under the age of 18.

3.3 Minimum wage laws

The federal minimum wage applies to high school students in the labor force, although with some

exceptions. The law currently requires that a minimum of 4.25 per hour be paid to employees

9

4 THE EFFECT OF COMPULSORY SCHOOLING ON EMPLOYMENT

under 20 years of age during the first 90 consecutive calendar days with an employer, after which

they should receive the regular federal minimum wage. Many high school students work as tipped

employees 4 in the hospitality industry and are subject to a lower minimum wage. The law however

requires the employer to pay the federal minimum wage in case the tipped minimum wage plus tips

is below the federal minimum wage .

4 The effect of compulsory schooling on employment

In this section I discuss three unambiguous theoretical predictions of the effect of compulsory

schooling and employment certificate legislation on employment. I imagine a simple model in

which employers make choices about which general groups of workers are best suited for a type

of job. The utility of filling a job with a certain type of worker depends on the skills the worker

brings, hiring and firing costs, minimum wage provisions, etc. Employers have prior beliefs about

the utility each type of worker would bring. Beliefs are updated after the hiring decision. Employers

may thus develop preferences for adult workers or for teenage workers, to identify just two poten-

tial categories. Hiring teenage workers may present disadvantages, such as low skills, competing

interests (school) that may impact effort on the job, burdensome child labor hiring provisions, low

flexibility in regards to hours and nightime work because of child labor provisions, higher proba-

bility of quitting, etc. Teenage workers may also bring advantages to employers, notably low wage

expectations, temporary subminimum wages for 90 days, low healthcare costs, etc. Because of the

low level of skills required in jobs filled by teenagers, substitutability is high, and we generally

expect each advantage and disadvantage of teenage workers relative to adults to weigh heavily in

the decision to hire teenagers.

Suppose also that teenagers make decision about employment and enrollment based on expected

utility, which depends on their taste for leisure, on the bureucratic cost of applying for jobs and

obtaining work permits , on the expected value of waged employment and on their valuation of

academic performance, which in turn may be a function of future expected earnings.

Hypothesis 1. The employment rate of sixteen and seventeen year old students is lower if an

employment certificate is required for employment. By extension, the employment rate of sixteen

and seventeen year old students is lower in states requiring such certificates.

4A tipped employee is any employee working in an occupation in which he or she regularly receives more than 30dollars a month in tips.

10

4 THE EFFECT OF COMPULSORY SCHOOLING ON EMPLOYMENT

In states that require employment certificates for sixteen and seventeen year olds, employers

have to pay a monetary cost (10-15 dollars) for the issuance of an employment certificate, and then

store the certificates for future inspections. The employers are at risk of being fined for improp-

erly complying with child labor legislation, and this risk increases if an employment certificate is

required. For students, obtaining a work permit discourages employment as it reduces the utility

of employment through all or some of the following: filling in forms, talking to and obtaining per-

mission from school counselors and parents, visiting the State Department of Labor, obtaining a

certificate of physical health, being at risk of losing work privileges if school performance drops.

Hypothesis 2. After the enactment of compulsory schooling laws, employers hire fewer sixteen

and seventeen year old students (just sixteen or both sixteen and seventeen, depending on the new

maximal school leaving age provided by the compulsory schooling reform).

Following a compulsory schooling policy, the first order effect is that fewer students apply for

jobs. Some of the students who would have dropped out may completely displace paid employment

for full time schooling and comply with the policy. Regardless of how many students decide to

substitute employment for enrollment, the total pool of job applicants is unambiguously lower if

any substitution at all is going on. We assume there are no ’deniers’, or students who would have

stayed in school without seeking a job but choose to seek employment simply because the compul-

sory schooling law is passed.

The second order effect is that employers may negatively update their beliefs about teenage

workers and choose to hire fewer teenagers overall. This is an example of negative selection: the

students who choose to apply for exceptions to compulsory schooling laws or simply break the law

and leave before the required age may be facing material hardship or may have very low academic

skills and struggle in school. Thus the pool of students applying for jobs while not enrolled in

school may be negatively selected.

Employers will also face greater uncertainty about the skills and motivations of students enrolled

in school following the compulsory schooling policy. Such laws may lead employers to change their

beliefs about the average attributes of enrolled students, since students who would have stayed in

school are mixed with students who would have dropped out. Thus compulsory schooling policies

increases the noise in the employer’s ability to assess skills, motivation and work ethic of high

school students and may induce employers to substitute high school students for other categories

of workers altogether.

To formalize the previous argument, consider a model in which employers have prior beliefs

11

4 THE EFFECT OF COMPULSORY SCHOOLING ON EMPLOYMENT

about categories of workers. For example, they believe the productivity V of job candidates follows

a distribution G(v, σ2), with different distributions for high school students, high school dropouts,

and adults. In a perfectly competitive market, firms would pay wages equal to expected worker

value. Employers however face different constraints for the three categories of workers: different

minimum wages for teenagers for 90 days, the cost of complying with child labor legislation, the

cost of issuing employment certificates, different healthcare costs depending on the employee’s age,

etc. The employer’s problem is to maximize profits by choosing a mix of candidates from the high

school student, high school dropout and adult pools, given the hiring and minimum pay constraints.

Suppose employers form their beliefs about productivity of high school students and high school

dropouts based on an enrollment signal, s. Let s be measure of perceived attachment to school-

ing, ranging from no further schooling to full enrollment, with intermediate categories : part-time

schooling, adult education, GED. Assume employers believe the productivity of full high school

attendance (s(1)), is higher than the productivity of students who have no further attachment to

schooling s(0)

The pool of minors will then have a distribution of productivity of G(v(S), σ2), where S is a

weighted average of schooling signals. S = ps(1)+(1−p)s(0), where p is the proportion of students

enrolled in school.

After the compulsory schooling reform, employers no longer attach the same value to the full

time school attendance signal, as they know would be dropouts are now a part of the pool of high

school students. The mixing of potential dropouts with students who would have stayed in school

is a very salient aspect in the debates surrounding compulsory schooling reforms. Opponents of

compulsory schooling often argue students forced to stay in school may have academic and be-

havioral problems that could contaminate the academic performance and discipline climate of the

school. Let δ denote the discount rate employers apply to the full time schooling signal after the

compulsory schooling law is passed.

Employers may also downgrade the value they attach to the high school dropout signal, by a

discount rate γ . This is because students who continue to drop out after the reform is passed

may be negatively selected. For example, the state of Kansas explicitly makes parents of students

looking to drop out before the age of 18 sign a disclaimer that the child lacks skills and his or her

earnings will be lower later in life. Other students may break the law and face the fines and risk

of imprisonment or other sanctions associated with habitual truancy. Some others may be coming

from low socio-economic backgrounds and have a demonstrable need to work.

12

5 DATA AND SUMMARY STATISTICS

After the compulsory schooling reform, the pool of minors will have a distribution of produc-

tivity of G(v(S′), σ2), where S’ is the weighted average of schooling signals after the compulsory

schooling reform. S′ = p′s(1)δ + (1 − p′)s(0)γ.

If S′ < S, employers will hire fewer minors under 18 and more employees over the age of

18, assuming the distribution of productivity is unchanged for adults. S’ is lower than S if

p′s(1)δ + (1 − p′)s(0)γ < ps(1) + (1 − p)s(0).

The perceived distribution of skills will be lower if the discount rates δ and γ are low, meaning

employers are more wary of the negative selection process, and when the enforcement of the law

is lax, meaning p′ − p is small. The model predicts the worst disemployment outcomes occur in

situations where the debate surrounding compulsory schooling laws is protracted, casting a negative

stigma on both dropouts and the average schoolstayer (low δ, γ), and the enforcement of the policy

ends up being lax (low p′ − p).

5 Data and summary statistics

The main data sources are the nationally representative Current Population Survey (CPS) March

Supplements 1962-2013 and the 1979-2014 NBER Merged Outgoing Rotation Group supplements.

The National Bureau of Economic Research extracts of the CPS outgoing rotation files record

the fourth and eighth (final) interview for households included in the sample. They provide in-

formation on usual weekly hours worked and earnings along other demographic and work related

items. Individuals are interviewed if they are aged 16 and over. Households in the CPS are inter-

viewed each month for four months, not contacted for eight months and then interviewed again for

four more months. By pooling yearly extracts we will find the same individuals interviewed twice

(in the fourth and eighth months). However, all analyses below are conducted with samples of

individuals of a certain age. By restricting the analysis to single age cohorts, we avoid the problem

of observing individuals twice in combined samples.

Information on compulsory schooling changes is compiled from state session laws and statutes

, corroborated with Monthly Labor Review updates on state labor laws, information in Whitfield

and Whitehurst (2012), Mackey and Duncan (2013) and email or phone conversations with State

Departments of Education. Information on employment certificate legislation is compiled from the

Department of Labor Wage and Hour Division website, as well as U.S. Department of Labor ,

State Child Labor standards publications. Information on minimum wage changes and state aver-

13

6 ESTIMATION METHODS

age yearly unemployment rates is also compiled from the US Department of Labor public data.

A series of covariates used in the analysis are constructed variables: using information on house-

holds, i sum weekly earnings of individuals aged 18 and over and divide total household weekly

earnings by the number of household members. Earnings per family member are updated to 2014

values using the Bureau of Labor statistics CPI inflation calculator. I then create deciles of house-

hold income per member. I also record the maximal educational attainment in each household

where a sixteen or seventeen year old is surveyed. The employment/enrollment status presented

in Table 3 is created using different variables that measure enrollment, labor force status and part

time/full time work status.

Table 3 presents the school enrollment and employment status of sixteen and seventeen year

olds during the school year for selected years. As expected, the most significant changes in employ-

ment and enrollment rates are associated with recessionary periods. The percentage of students

reporting being unemployed or not in the labor force has actually dropped over the past 25 years,

even during recessions. The unemployed are however a larger share of those in the labor force,

since the labor force participation rate has dropped considerably and there has been a substantial

increase in teenagers reporting being enrolled and not in the labor force.

Table 4 compares the employment status of sixteen and seventeen year olds in states that require

employment certificates past the age of sixteen versus states that do not have such a requirement.

Fewer high school students are working part time in states that require employment certificates:

10.96 per cent, compared to 14.24 per cent in states that do not have such requirements. In states

requiring certificates, students are more likely to be enrolled in full time schooling and not working.

For example, 62.78 per cent of seventeen year olds are enrolled in school and not working in states

requiring certificates, compared to 68.3 per cent in states with no mandatory certificates.

6 Estimation methods

I estimate the effect of compulsory schooling laws, state requirements for employment certificates

for minors above the age of 16, and minimum wages on school enrollment and employment, at ages

16 and 17, separately. I split the analysis for different ages because such a split is relevant for

the way compulsory schooling laws are enforced. Also, such a split avoids observing the same set

of individuals twice, in different years, as individuals are observed in the CPS data in the fourth

month of their interview and one year later, in the eight month in which the household is observed.

14

7 RESULTS

To estimate the effect of compulsory schooling laws on school enrollment, I regress an indicator

variable for whether the student was enrolled in school during the school year (September through

May) on indicators for whether the student faced a minimal school leaving age of seventeen or eigh-

teen in their state of residence at the time when they were sixteen years old. The regression controls

for the log of the state average minimum wage, the state yearly average unemployment rate as well

as for a vector of demographic characteristics. The individual controls Xist include: gender, race,

an ethnicity indicator variable (Hispanic), indicator variables for the highest level of educational

attainment in the household in which the teenager is surveyed and a measure of whether the area of

residence was an outlying metropolitan region or a rural area (the baseline being individuals living

in principal cities of metropolitan areas). Regressions include state (δs) and year (δt) fixed effects

and errors are clustered at the state level. Regressions are run using individual observations but

also by collapsing observations into state-year averages. This specification is also run separately

for teenagers living in households in the lower decile of household earnings per member.

ENROLLEDist = Xist+DROPAGE17st+DROPAGE18st+lnMWst+UNEMPLst+δs+δt+εist

To measure employment outcomes, I create an indicator variable that takes on the value one if

the sixteen year old or, in another specification, the seventeen year old is working during the school

year. I regress this variable on the previous set of controls.

EMPLOY EDist = Xist+DROPAGE17st+DROPAGE18st+lnMWst+UNEMPLst+δs+δt+εist

To assess the effect of employment certificate requirements, i use the following specification:

EMPLOY EDist = Xist + CERTIFICATEst + lnMWst + UNEMPLst + δs + δt + εist

, where the CERTIFICATE variable measures whether states had a law in place requiring

employment certificates for sixteen and seventeen year olds.

7 Results

Changes in the minimal school leaving age have a small and statistically insignificant effect on

the overall enrollment of sixteen or seventeen year olds. However, the effect of these policies is

larger and statistically significant for individuals in the bottom decile of household income. For

these students, compulsory schooling laws increase the enrollment rate by about three percentage

points. The higher effect of compulsory schooling for students from low income households is consis-

tent with models that predict higher dropout rates for students from credit constrained households.

15

7 RESULTS

Laws requiring a higher minimal school leaving age have a negative effect on the employment

rate. Requiring a minimal school leaving age of 18 reduces the employment rate at age 16 by 2.1

percentage points and by 1.7 percentage points at age 17 (Table 6, for years 1979-2014). Minimum

wage laws have a large and statistically significant negative effect on the employment rate. It is

interesting to note that if the increases in the minimum school leaving age reduce the probability

that high school students work during high school, and these changes are contemporaneous with

minimum wage increases, the disemployment effects of the latter will be confounded and upwardly

biased. Out of the 20 states that increased compulsory schooling laws since 1987, eleven did so at

the same time as increasing minimum wages. For example , a regression that omits the compulsory

schooling indicators shown in Table 6 will show a coefficient of -.011 (p value 0.001), larger than

the coefficient obtained in a regression that accounts for compulsory schooling laws (-0.099, p value

0.018).

Three of the twenty compulsory schooling changes occured during the recent recession. We

may be concerned that the drastic decrease in employment during the recent recession may bias

results. Table 6 also presents estimates for the effect of compulsory schooling laws on employment

between 1992 and 2008. The effect is a negative 2.5 percentage point decrease in employment for

16-year-olds for laws requiring schooling until 17 and a decrease of 1.9 per cent for 17-year-olds,

both statistically significant.

To test the robustness of the effects of compulsory schooling laws on the employment rate, I

estimate the effect of minimal age increases at ages where such policies should not have had an

effect. Table 7 presents these estimates, showing that effects are statistically significant at ages

16 and 17, as expected under Hypothesis 2, and not statistically significant at other ages, with

the exception of a coefficient in the regression for the sample of 18 year olds. The weak statistical

significance of that coefficient may be an artifact of repeated testing but it may also be related to

the compulsory schooling law if employers negatively update their belief about all teen workers at

the time of the compulsory schooling increase, as suggested by Hypothesis 2.

The effect of relaxing employment certificate requirements is analyzed in Table 8. I estimate

the effect of dropping work permit requirements for five states that did so between 1977 and 1995.

Dropping the cerfificate requirement resulted in a 2.7 percentage point increase in the employ-

ment rate of 16-year-olds and a 1.9 percentage point increase for 17 year olds, both statistically

significant. These results should not be interpreted as the causal effect of employment certificate

requirements for sixteen and seventeen year olds, since laws to relax such requirements in some

cases also relaxed nightwork and total hours restrictions for sixteen and seventeen year olds. The

effects of the employment certificate requirement are thus confounded with other child labor pro-

16

8 CONCLUSIONS

visions in the specification in Table 8.

8 Conclusions

Increases in the minimal school leaving age in 20 states since 1987 account for a large share of the

”unaccounted” secular decline in teenage employment. Previous macroeconomic estimates placed

the unaccounted component at between two and three percentage points (Smith, 2011). Given the

two percentage point decrease in employment in the 20 states implementing compulsory schooling

policies, and the fact that these states represent 44 per cent of the US population, I estimate com-

pulsory schooling laws can account for about one percentage point out of the national level decline

in teenage employment.

While potentially raising high school dropouts’ chances of finishing high school, compulsory

schooling laws may also reduce exposure to the skills, earnings and hands on experience acquired

while employed. The present analysis indicates that employment will decrease not only for potential

high school dropouts who are kept in school longer, but also for other students, as employers may

displace high school students with workers of other ages after a compulsory schooling increase. The

large and positive effects that relaxing work permit legislation had on employment also highlight

the need for minimal school leaving age increases to be analyzed in conjunction with employment

certificate legislation.

17

9 REFERENCES

9 References

Aaronson, Daniel, Kyung-Hong Park, and Daniel Sullivan. 2006. The decline in teen labor force

participation. In Federal Reserve Bank of Chicago economic perspectives, 2006 Q1, 218.

Acemoglu, D., and J. Angrist. 2001. How large are human capital externalities? Evidence from

compulsory schooling laws. NBER Macroannual: 9 59.

Allegretto, Sylvia, Dube, Arindrajit, Reich, Michael, 2011. Do Minimum Wages Really Reduce

Teen Employment? Accounting for Heterogeneity and Selectivity in State Panel Data, Industrial

Relations, Vol. 50, No. 2 (April 2011).

Angrist, J. D., and A. Krueger. 1991. Does compulsory school attendance affect schooling and

earnings? Quarterly Journal of Economics 106 (4): 979 1014.

Eckstein, Zvi, and Wolpin, Kenneth I. Youth Employment and Academic Performance in High

School. Centre for Economic Policy Research Discussion Paper no. 1861. London: Centre for

Economic Policy Research, 1998

Lewin-Epstein, N. 1981. Youth Employment During High School. Chicago, IL: National Center

for Education Statistics.

Mackey, P. E., Duncan, T. G. (2013). Does raising the state compulsory school attendance age

achieve the intended outcomes? (REL 2014005). Washington, DC: U.S. Department of

Education, Institute of Education Sciences, National Center for Education Evaluation and

Regional Assistance, Regional Educational Laboratory Mid-Atlantic. Retrieved from

http://ies.ed.gov/ncee/edlabs

Neumark, David and William Wascher. 2006. ”Minimum Wages and Employment: A Review of

Evidence from the New Minimum Wage Research.” National Bureau of Economic Research

Working Paper 12663. Cambridge, MA: National Bureau of Economic Research.

http://www.nber.org/papers/w12663

Oreopoulos, Philip. 2009. Would More Compulsory Schooling Help Disadvantaged Youth?

Evidence From Recent Changes to School- Leaving Laws in Gruber, Jonathan [ed], The Problems

of Disadvantaged Youth: An Economic Perspective , University of Chicago Press,. pp. 85 112

Smith, Christopher L, 2011. ”Polarization, Immigration, Education: What’s Behind the Dramatic

Decline in Youth Employment?” Federal Reserve, October 2011.

http://www.federalreserve.gov/pubs/feds/2011/201141/201141pap.pdf

18

10 APPENDIX

Smith, Christopher L, 2012. The Impact of Low-Skilled Immigration on the Youth Labor Market,

Journal of Labor Economics Vol. 30, No. 1 (January 2012), pp. 55-89

Tyler, John, 2003. Using State Child Labor Laws to Identify the Effect of SchoolYear Work on

High School Achievement , Journal of Labor Economics, Vol. 21, No. 2 (April 2003), pp. 381-408

Whitehurst, Grover J, Whitfield, Sarah, 2012. Compulsory school attendance. What research

says and what it means for state policy , The Brookings Institution, available at

www.brookings.edu/research/papers [Accessed April 15, 2014]

10 Appendix

19

10 APPENDIX

Figure 1: Employment rate for sixteen and seventeen year olds who were enrolled during the schoolyear

Source: Current Population Survey weighted data, 1984-2014. Observations only include individuals reportingbeing enrolled in school and being employed , September through May. Shaded regions indicate years with National

Bureau of Economic Research designated recessions.

Figure 2: Enrollment rate for sixteen and seventeen year olds, September-May averages

Source: Current Population Survey weighted data, 1984-2014. Observations only include individuals reportingbeing enrolled in school and being employed , September through May. Shaded regions indicate years with National

Bureau of Economic Research designated recessions.

20

10 APPENDIX

Figure 3: Employment rate for different age groups (September-May averages)

Source: Current Population Survey weighted data, 1984-2013. Observations only include individuals reportingbeing employed September through May. Shaded regions indicate years with National Bureau of Economic

Research designated recessions.

21

10 APPENDIX

Table 1: State minimal school leaving ages, selected years

State 1990 2000 2014 State 1990 2000 2014

Alabama 16 16 17

Montana 16 16 16

Alaska 16 16 16

Nebraska 16 16 18

Arizona 16 16 16

Nevada 17 17 18

Arkansas 17 18 18

New Hampshire 16 16 18

California 18 18 18

New Jersey 16 16 16

Colorado 16 16 17

New Mexico 18 18 18

Connecticut 16 16 18

New York 16 16 16

Delaware 16 16 16

North Carolina 16 16 16

DC 17 18 18

North Dakota 16 16 16

Florida 16 16 16

Ohio 18 18 18

Georgia 16 16 16

Oklahoma 18 18 18

Hawaii 18 18 18

Oregon 18 18 18

Idaho 16 16 16

Pennsylvania 17 17 17

Illinois 16 16 17

Rhode Island 16 16 18

Indiana 16 16 18

South Carolina 16 17 17

Iowa 16 16 16

South Dakota 16 16 18

Kansas 16 18 18

Tennessee 17 17 17

Kentucky 16 16 16

Texas 17 17 18

Louisiana 17 17 18

Utah 18 18 18

Maine 17 17 17

Vermont 16 16 16

Maryland 16 16 16

Virginia 18 18 18

Massachussetts 16 16 16

Washington 18 18 18

Michigan 16 16 16

West Virginia 16 16 16

Minnesota 16 16 16

Wisconsin 18 18 18

Mississippi 16 17 17

Wyoming 16 16 16

Missouri 16 16 17

Source: Author’s review of state session laws and statutes , corroborated with Monthly Labor Review

updates on state labor laws, Whitehurst and Whitfield (2012), Mackey and Duncan (2013) and email or phone

conversations with State Departments of Education.

22

10 APPENDIX

Figure 4: Minimal school leaving ages in 2014

Map created with Statplanet software, replicates information in Table 1, column ”2014”

Figure 5: Employment certificate requirements in 2014

Map created with Statplanet software ,replicates information in Table 2

23

10 APPENDIX

Table 2: State requirements for employment certificates for minors, 2003-2014

Notes: 1. Nevada requires an employment certificate only for minors under the age of 14. 2. During school hours.3. Only if not enrolled in school. 4. Only for minors enrolled in school. 5. 16 and 17 during school term. Source: Information on state certificate requirements comes from the Department of Labor Wage and Hour division website, http://www.dol.gov/whd/state/certification.htm#5

No requirement Arizona, Florida, Idaho, Kentucky, Montana, Nevada1, Oregon, South Carolina, South Dakota, Tennessee, Texas, Utah, Wyoming

Employment certificate required for minors under the age of 16

Arkansas, Colorado2, Connecticut, Hawaii, Illinois, Iowa, Kansas3, Maine, Minnesota2, Mississippi, Missouri, Nebraska, New Hampshire, New Mexico, North Dakota, Oklahoma, Rhode Island, Vermont, Virginia, West Virginia.

Employment certificate required for minors under the age of 17

Alaska

Employment certificate required for minors under the age of 18

Alabama, California4, Delaware, District of Columbia, Georgia, Indiana, Louisiana, Maryland, Massachussetts ,Michigan, New Jersey, New York, North Carolina, Pennsylvania, Ohio5 , Washington, Wisconsin.

24

10 APPENDIX

Table 3: Employment/ enrollment status of 16 and 17 year olds, selected years

Source: NBER extracts of the Merged Outgoing Rotation Files of the Current Population Survey, 1989-2014. Theproportions are calculated using population weights. Averages for the months of September through May.

25

10 APPENDIX

Table 4: Employment/enrollment status of minors by type of state mandated employment certifi-cate

Source: Data come from the NBER extracts of the Merged Outgoing Rotation Files of the Current Population Survey, 2003-2012. Information on state certificate requirements comes from the Department of Labor Wage and Hour division website, http://www.dol.gov/whd/state/certification.htm. Notes: The school year excludes the months of June, July and August. For information on states included in the no requirement/requiring certificates categories, see Table 3.

Employment status at age 16 (during the school year)

Employment status at age 17 (during the school year)

States with no

requirement

States requiring

certificates

States with no

requirement

States requiring

certificates

% of total

Full time school only 77.04 81.00

62.78 68.30

Full time school, part time work 14.24 10.96

23.90 19.74

In school part time, working part time 0.16 0.15

0.55 0.41

Working full time, in school full time 0.50 0.29

0.90 0.68

Working full time, in school part time 0.04 0.05

0.09 0.1

Unemployed, in school full time 4.43 4.14

4.64 4.47

Unemployed, in school part time 0.08 0.06

0.21 0.17

Not enrolled, working full time 0.41 0.49

1.19 0.91

Not enrolled, working part time 0.56 0.45

1.34 1.01

Not enrolled, unemployed 0.31 0.27

1.08 0.97

Not enrolled, not in labor force 2.21 2.13

3.30 3.24

Number of observations 30,356 18,899

30,074 18,613

26

10 APPENDIX

Table 5: Regression estimates for the probability of enrollment, ages 16 and 17

1

(1) (2) (3) (4) Enrollment

rate at age 16 Enrollment rate at age 16, bottom decile of

HH income

Enrollment rate at age 17

Enrollment rate at age 17, bottom decile of

HH income

Must stay in school until 17 .001 .010 (.005) (.012) Must stay in school until 18 .005 .031* .007 .037* (0.003) (.015) (0.007) (.015) Cons

.916***

.840***

.862***

.846***

(.032) (.042) (.052) (.085)

State fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes N 1,213 1,213 1,213 1,213 Notes: OLS regressions using Current Population Survey merged outgoing rotation groups(MORG) data for 1991–2014. Errors are clustered at the state level. Clustered robust standard errors in parantheses, + 0.10 signified p<0.1; * Signifies p < 0.05; ** signifies p < 0.01; ***

signifiesp <0.001. Regressions include controls for %female, minimum wages, unemployment rate, %Black, %Hispanic, % Female, %Rural, %Outside principal city.

27

10 APPENDIX

Tab

le6:

Reg

ress

ion

esti

mat

esfo

rth

ep

rob

abil

ity

ofem

plo

ym

ent,

ages

16an

d17

E

mp

loym

ent

rate

at

age

16,

Em

plo

ymen

t ra

te a

t ag

e

17

1979-2

014

1992-2

008

1979-2

014

1992-2

008

Must

sta

y in

sch

oo

l un

til 17

-.

018*

(0

.007)

-.

025*

(.

011)

M

ust

sta

y in

sch

oo

l un

til 18

-.021**

(.

008)

-.019

(.012)

-.017*

(.

007)

-.019

+

(.011)

L

n M

inim

um

Wag

e

-.099*

-.

099*

-.

069*

-.

060

+

(.

040)

(.039)

(.029)

(.034)

Un

emp

loym

ent

rate

-.

016**

* -.

023**

* -.

021**

* -.

022**

*

(.001)

(0.0

04)

(0.0

01)

(.003)

C

on

s .3

87**

* .4

49**

* .5

40**

* .4

93**

*

(.073)

(.069)

(.061)

(.078)

Sta

te fix

ed e

ffect

s

Yes

Y

es

Yes

Y

es

Yea

r fix

ed e

ffect

s Y

es

Yes

Y

es

Yes

N

1,8

21

866

1,8

21

866

No

tes:

OL

S r

egre

ssio

ns

usi

ng

Curr

ent

Po

pula

tio

n S

urv

ey m

erge

d o

utg

oin

g ro

tati

on

gro

up

s(M

OR

G)

dat

a fo

r ye

ars

ind

icat

ed E

rro

rs a

re c

lust

ered

at

the

stat

e le

vel

. C

lust

ered

ro

bust

sta

nd

ard

err

ors

in

par

anth

eses

, +

0.1

0 s

ign

ifie

d p

<0.1

; * S

ign

ifie

s p

< 0

.05;

** s

ign

ifie

s p

< 0

.01

; *** s

ign

ifie

sp <

0.0

01. R

egre

ssio

ns

incl

ud

e co

ntr

ols

fo

rn

%B

lack

, %

His

pan

ic, %

Fem

ale,

%R

ura

l, %

Outs

ide

pri

nci

pal

cit

y, %

Of

ho

use

ho

lds

wh

ere

the

hig

hes

t le

vel

of

educa

tio

nal

att

ain

men

t is

So

me

colle

ge, %

Of

ho

use

ho

lds

wh

ere

the

hig

hes

t le

vel

of

educa

tio

nal

att

ain

men

t is

co

llege

or

grad

uat

e d

egre

es.

28

10 APPENDIX

Table 7: Robustness test for effects of compulsory schooling policies on the employment rate:unaffected age groups

Employment rate

Age 16 Age 17 Age 18 Age 19 Age 20 Age 21 Age 22 Age 23 Age 24

Maximal leaving age 17

-.018* (0.007)

-.007 (.009)

-0.019* (0.007)

-.006 (0.006)

-.005 (0.007)

.000 (0.009)

.002 (0.007)

.011 (.008)

-.003 (0.007)

Maximal leaving age 18

-.021** (.008)

-.017* (.007)

-0.008 (0.007)

-.001 (0.007)

-.002 (0.009)

0.001 (0.012)

-.012 (0.009)

-.006 (0.006)

-0.013+ (0.007)

N 1,821 1,821 1,821 1,821 1,821 1,821 1,821 1,821 1,821

Notes: OLS regressions using Current Population Survey merged outgoing rotation groups(MORG) data for 1979-2014.

Errors are clustered at the state level. Clustered robust standard errors in parantheses, + 0.10 signified p<0.1; * Signifies p

< 0.05; ** signifies p < 0.01; *** signifiesp <0.001. Regressions include controls forn %Black, %Hispanic, % Female, %Rural,

%Outside principal city, %Of households where the highest level of educational attainment is Some college, %Of

households where the highest level of educational attainment is college or graduate degrees.

29

10 APPENDIX

Table 8: Effect of work permit requirements on the employment rate of high school students

Employment rate at age 16,

1977-2002

Employment rate at age 17,

1977-2002

Certificate requirement until age 18 is dropped

.027** (.009)

.019* (.009)

Ln Minimum Wage

.036

.077

(.090) (.086)

Unemployment rate -.011*** -.019*** (0.002) (.000) Cons .286*** .441** (.058) (.101)

State fixed effects Yes Yes Year fixed effects Yes Yes N 711 711 Notes: OLS regressions using Current Population Survey March data, 1977-2002 and includes individuals who were enrolled in high school. Data is collapsed at the state-year level. Errors are clustered at the state level. Clustered robust standard errors in parantheses, + 0.10 signified p<0.1; * Signifies p < 0.05; ** signifies p < 0.01; *** signifies p <0.001.

30